When you're dealing with a plasma at 20 million kelvin, the leftovers from that "significant fraction" that don't get reflected can really ruin your day...
LLNL scientists almost tested a similar device (see 'MFTF') but the project was cancelled in 1986 before it was turned on.
It is possible though that though earlier attempts failed because the computational resources and the plasma dynamics modeling available at the time were not sufficient to design such a reactor.
I would guess that we have a much better understanding of plasma physics and faaaaar greater ability to model it now - that could lead to designs that can mitigate previously discovered issues with the concept.
LLNL scientists almost tested a similar device (see 'MFTF') but the project was cancelled in 1986 before it was turned on.